Non-informative priors do not exist A dialogue with José M. Bernardo

Irony, Telba; Singpurwalla, Nozer D.

doi:10.1016/s0378-3758(97)00074-8

Cited by 86 publications

(48 citation statements)

References 44 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This can be a potential advantage, but in many situations there is a desire for the 'data to dominate' when no prior information is available (or when MCMC methods are being used for computational convenience and the researcher does not want to include prior information), which has led to the use of vague or reference priors [15]. We do not advocate the use of the term non-informative prior distribution as we consider all priors to contribute some information [16][17][18]. If data is sparse then even prior distributions that are intended to be vague may exert an unintentionally large degree of in uence on any inferences.…”

Section: Introductionmentioning

confidence: 99%

How vague is vague? A simulation study of the impact of the use of vague prior distributions in MCMC using WinBUGS

et al. 2005

View full text Add to dashboard Cite

There has been a recent growth in the use of Bayesian methods in medical research. The main reasons for this are the development of computer intensive simulation based methods such as Markov chain Monte Carlo (MCMC), increases in computing power and the introduction of powerful software such as WinBUGS. This has enabled increasingly complex models to be fitted. The ability to fit these complex models has led to MCMC methods being used as a convenient tool by frequentists, who may have no desire to be fully Bayesian. Often researchers want 'the data to dominate' when there is no prior information and thus attempt to use vague prior distributions. However, with small amounts of data the use of vague priors can be problematic. The results are potentially sensitive to the choice of prior distribution. In general there are fewer problems with location parameters. The main problem is with scale parameters. With scale parameters, not only does one have to decide the distributional form of the prior distribution, but also whether to put the prior distribution on the variance, standard deviation or precision. We have conducted a simulation study comparing the effects of 13 different prior distributions for the scale parameter on simulated random effects meta-analysis data. We varied the number of studies (5, 10 and 30) and compared three different between-study variances to give nine different simulation scenarios. One thousand data sets were generated for each scenario and each data set was analysed using the 13 different prior distributions. The frequentist properties of bias and coverage were investigated for the between-study variance and the effect size. The choice of prior distribution was crucial when there were just five studies. There was a large variation in the estimates of the between-study variance for the 13 different prior distributions. With a large number of studies the choice of prior distribution was less important. The effect size estimated was not biased, but the precision with which it was estimated varied with the choice of prior distribution leading to varying coverage intervals and, potentially, to different statistical inferences. Again there was less of a problem with a larger number of studies. There is a particular problem if the between-study variance is close to the boundary at zero, as MCMC results tend to produce upwardly biased estimates of the between-study variance, particularly if inferences are based on the posterior mean. The choice of 'vague' prior distribution can lead to a marked variation in results, particularly in small studies. Sensitivity to the choice of prior distribution should always be assessed.

show abstract

Section: Introductionmentioning

confidence: 99%

How vague is vague? A simulation study of the impact of the use of vague prior distributions in MCMC using WinBUGS

et al. 2005

View full text Add to dashboard Cite

show abstract

“…However, uninformative priors do not really exist (see in [9]) and all priors are informative in some ways. Moreover, there have been various names associated with uninformative priors including diffuse, minimal, non-informative, objective, reference, uniform, vague, and perhaps weakly informative etc.…”

Section: Uninformative Priormentioning

confidence: 99%

Bayesian Inference Application

Kumam¹,

Saipara²

2017

Bayesian Inference

View full text Add to dashboard Cite

show abstract

“…Different interpretations and justifications of non-informative (traditionally called 'uninformative') priors have been suggested over the years, including invariance -for example, [30,31]. Nowadays, however, uninformative priors do not actually exist, because all priors are informative in some way [32]. Proper distributions with inverse gamma such as 0.001 for BUGS modelling were presented by [25,26].…”

Section: Figure 1: Visual Layout Of the Recorded Uncertain Variablesmentioning

confidence: 99%

Production Uncertainties Modelling by Bayesian Inference Using Gibbs Sampling

Azizi

Ali

Ping

et al. 2015

SAJIE

View full text Add to dashboard Cite

Analysis by modelling production throughput is an efficient way to provide information for production decision-making. Observation and investigation based on a real-life tile production line revealed that the five main uncertain variables are demand rate, breakdown time, scrap rate, setup time, and lead time. The volatile nature of these random variables was observed over a specific period of 104 weeks. The processes were sequential and multi-stage. These five uncertain variables of production were modelled to reflect the performance of overall production by applying Bayesian inference using Gibbs sampling. The application of Bayesian inference for handling production uncertainties showed a robust model with 2.5 per cent mean absolute percentage error. It is recommended to consider the five main uncertain variables that are introduced in this study for production decision-making. The study proposes the use of Bayesian inference for superior accuracy in production decision-making. OPSOMMINGAnalise deur middel van die modellering van produksie deurset is 'n effektiewe manier om inligting vir produksiebesluitneming te verskaf. Die waarneem en ondersoek van 'n teëlproduksielyn het getoon dat die vyf hoof onsekerheidsveranderlikes die vraagtempo, breektyd, skraptempo, opsteltyd en leityd is. Die vlugtige aard van hierdie toevalsveranderlikes is waargeneem oor 'n tydperk van 104 weke. Die prosesse was opeenvolgend en multi-stadium. Die vyf onsekerheidsveranderlikes van produksie is gemodelleer om die algehele vertoning van die produksie te weerspieël deur gebruik te maak van Bayesiese afleiding met Gibbs monsterneming. Die toepassing van Bayesiese afleiding vir die hanteer van produksie onsekerhede het 'n robuuste model, met 'n tweeen-'n-half persent gemiddelde absolute persentasie fout, tot gevolg gehad. Dit word aanbeveel dat die vyf belangrikste onsekerheidsveranderlikes, wat in hierdie studie bekendgestel is, oorweeg moet word vir produksiebesluitneming. Die studie stel die gebruik van die Bayesiese afleiding tegniek voor om sodoende beter akkuraatheid in produksiebesluitneming te verkry.

show abstract

Non-informative priors do not exist A dialogue with José M. Bernardo

Cited by 86 publications

References 44 publications

How vague is vague? A simulation study of the impact of the use of vague prior distributions in MCMC using WinBUGS

How vague is vague? A simulation study of the impact of the use of vague prior distributions in MCMC using WinBUGS

Bayesian Inference Application

Production Uncertainties Modelling by Bayesian Inference Using Gibbs Sampling

Contact Info

Product

Resources

About